The new model is a convolutional neural network (CNN). That is, the algorithm imposes a filter
During development, scientists used millionsExamples of images from two Copernicus Sentinel-2 satellites operated by the European Space Agency (ESA). These satellites record every location on Earth every five days with a resolution of 10 x 10 meters per pixel. The result is the highest quality images currently available.
To calculate the height of the trees, data obtained during the NASA GEDI mission was used.
“Mission GEDI provides globaldistributed disparate data on the height of vegetation between 51 degrees north and south latitude, so in the process of learning the computer sees many different types of vegetation, ”explains one of the researchers.
The resulting model is able to automatically estimate vegetation height from over 250,000 images (about 160 TB of data).
The high efficiency of the neural network is also associated withThe fact that the researchers used not one CNN, but five. They trained independently, and each of them made its own estimate of the height of the tree. The model also takes into account the uncertainty in the data itself: for example, if the satellite image is hazy, the uncertainty is higher than under good atmospheric conditions.
“An important aspect for us was to informusers about estimation uncertainty,” says Lang, one of the researchers. - If all models match, then the answer is clear based on the training data. If the models come to different answers, it means there is more uncertainty in the estimate.”
Global map obtained using a neural networkthe height of the plant canopy will help ecologists monitor climate change. It may also be of interest to governments and administrative bodies.
“With the Sentinel-2, vegetation heights can be recalculated every five days, allowing you to track deforestation in the rainforest,” says Lang.
Read more:
It has been hunted for centuries: what do we know about the planet Vulcan next to the Sun
Physicists have experimentally confirmed a new fundamental law for liquids
Astronomers have found the source of mysterious radio bursts that come from space